Comparison,analysis And Application Of Remote Sensing Inversion Algorithms Of Photosynthetically Active Radiation

Photosynthetically active radiation(PAR)is not only a key factor measuring photosynthesis but also a driving factor of global change,which plays an important role in many research fields such as meteorology,agriculture and ecology.Current mainstream methods of estimating PAR by remote sensing can be categorized into three types:atmospheric radiative transfer model(ARTM),parameterized model based and look up table(LUT)-based methods.The LUT in the LUT-based methods is generally built based on ARTM.Although the PAR remote sensing inversion researches have make great progress in recent years,there are still many problems existing in current researches:there are a little quantitative evaluations and comparisons for different models;many models and methods can't effectively estimate direct radiation and diffuse radiation;current methods simply neglect the impact of cloud and topography.These existing problems seriously affect the estimation accuracy and applicability.To solve these problems,the aim of this paper is:compare and analyze the different performances among ARTM,parameterized model,and LUT-based methods,under all-skies conditions,from different aspects:estimation accuracy,distinguishing the direct and diffuse radiations,computational efficiency,etc;The more effective method with a better accuracy under all-skies in rugged terrain is used to estimate the PAR in Chongqing;finally,the estimated PARs in Chongqing are validated first and then its temporal and spatial distribution characteristics are analyzed.To achieve above objectives,the work performed in this study and the corresponding conclusions are as follows:1)After the comparison between SBDART and MODTRAN,it is concluded that:under clear skies,both two models are in good agreements with the in situ data,their errors are mainly caused by AOD,and the differences between their total radiation is caused by the difference between the direct radiation;under cloudy skies,SBDART obviously has a more performance in estimated accuracy than MODTRAN.In addition,SBDART has a high computational efficiency than MODTRAN.2)Through the comparison and analysis between dual-band parameterized models and broad-band parameterized models,it demonstrates that,under clear skies,dual-band model has a better performance in estimating diffuse and total radiation than broad-band model,while there is an opposite conclusion for the direct radiation;under cloudy skies,both two parameterized models have low accuracy against in situ data,and their accuracy are considerably lower than the accuracy of ARTMs.3)Through the comparison and analysis between the LUT-based methods,it demonstrates that,under all-skies,the LUT-based methods agree better with in situ data than parameterized models,while they have lower accuracy than the ARTMs.But the calculation speed of the LUT-based models is faster than the ones of ARTMs.In general,LUT-based method based on SBDART shows better performance than that of MODTRAN.4)The LUT-based models based on SBDART is used to retrieve PARs in Chongqing.Then the estimated PARs are validated against in situ data.Finally,the temporal and spatial distribution characteristics of PAR in Chongqing are analyzed and discussed from three aspects:spatial patterns,vertical distributions,and seasonal variations.The results show that the retrieved PAR has great consistence with in situ PAR.The R~2,MBE,and RMSE of PAR are 0.73,30.3 W/m~2,and 23.63 W/m~2 respectively.The tempo-spatial distribution of PAR in Chongqing is obvious.Cloud,terrain and the periodic variation of direct solar radiation are key issues of the tempo-spatial distribution of PAR.